#mathematic Imagine you want to cook a chicken, you will use the oven and want to cook the best chicken possible. You will put ingredients on it, salt, pepper, onions ... and put it in the oven at some temperature. Optimization allow you to select the best ingredients and temperature to cook the best chicken. ### Evolutionary algorithms Just imagine you have a bunch of people ready to test your chicken and a bunch of ingredients and oven. You will cook with random "settings" a batch of chickens and ask for feedback from the testers, this feedback is often called "fitness function", how good is the chicken. Genetic algorithm will be used as follow: take the tastiest chicken settings and reproduce the experience based on these settings with a slight bit of randomness added (mutations). After many iterations, you will have the best settings to cook your chicken ! ![[Chicken.png]] ### Gradient descent Evolutionary algorithms are nice and used a lot, they converge quickly, sometimes too quickly. [[Mathematic/Calculus/Automatic differentiation]] TODO next